The implementation of empathy to interactions with healthcare chatbots is considered promising since it is a crucial element in doctor-patient relationships. Existing research commonly conceptualizes the multifaceted concept of empathy unidimensional resulting in an insufficient consideration of different types of empathic expressions. Drawing on the assumption that emotional empathic expressions are perceived as more human-like but less authentic than empathic caring behavior, we compare the outcomes of both empathy types. We therefore conducted an online experiment with n=403 participants who interacted with one of five diagnosis chatbots each expressing a different type of empathy. First results support our hypothesis that emotional expressions increase social presence but decrease authenticity while the opposite is true for caring chatbots. Mediation analysis reveals that both function as opposing mediators between empathy type and the chatbot's likeability. Our research thus contributes to a differentiated view on empathy in interactions with chatbots.